Journal of Physiology (1988), 404, pp. 199-213 199With 7 text-figures

Printed in Great Britain

PHOTOCURRENTS OF CONE PHOTORECEPTORS OF THEGOLDEN-MANTLED GROUND SQUIRREL

BY TIMOTHY W. KRAFTFrom the Department of Neurobiology, Stanford University Medical Center,

Stanford, CA 94305-5401, U.S.A.

(Received 1 February 1988)

SUMMARY

1. Visual transduction in photoreceptors of the ground squirrel, Citellus lateralis,was studied by recording membrane current from individual cones in small pieces ofretina.

2. Brief flashes of light produced transient reductions of the dark current;saturating response amplitudes were up to 67 pA. A flash strength of about 11000photons /tm-2 at A,,. was required to give a half-saturating response. The stimulus-response relation was well fitted by an exponential saturation curve. Responsesbelow 20% of maximum behaved linearly.

3. The response to a dim flash in most cells had a time to peak of 20-30 ms andresembled the impulse response of a series of five low-pass filters.

4. The variance of the dim-flash response amplitude put an upper limit of 80 fA onthe size of the single photon response. Estimates based on the effective collecting areasuggest the single photon response to be of the order of 10 fA.

5. Flash responses of squirrel cones usually lacked the undershoot observed inprimate cones, although in about I of the cells a small undershoot developed duringrecording.

6. Background lights slightly shortened the time to peak of the flash response andreduced the integration time.

7. Spectral sensitivity measurements showed two classes of cones with peaksensitivities at about 520 and 435 nm. Rod sensitivity peaked near 500 nm. Spectralunivariance was obeyed by all three classes of cells.

8. The shapes of the spectral sensitivity curves of the rod and both types of coneswere similar to each other when plotted on a log wave number scale, but differedsignificantly from similar plots of monkey and human cone spectra.

9. The kinetics and sensitivity of flash responses of the blue- and green-sensitivecones were indistinguishable.

INTRODUCTION

The introduction of the suction electrode technique (Baylor, Lamb & Yau, 1979a)has allowed the study of relatively small photoreceptors including those frommonkey and human retinas (Baylor, Nunn & Schnapf, 1984, 1987; Schnapf, Kraft &

Baylor, 1987). One of the difficulties in studying cone transduction in primates is thatoutside the fovea, which comprises less than 1 % of the retinal surface, the cones arehidden within dense forests of rods. An alternative preparation is provided by theretina of the diurnal ground squirrel in which about 90% of photoreceptors are cones(Walls, 1942; West & Dowling, 1975; Jacobs, Fisher, Anderson, & Silverman, 1976;Long & Fisher, 1983). Ground squirrels are dichromatic and lack red cones, asdemonstrated by behavioural (Crescitelli & Pollack, 1972; Jacobs, 1978) andphysiological experiments (Tansley, Copenhaver & Gunkel, 1961; Michael, 1968; Gur& Purple, 1978; Raisanen & Dawis, 1983; Ahnelt, 1985; Jacobs, Neitz, & Crognale,1985). Thus this retina offers a simple colour system and greater access to the blue-sensitive cones.

This paper describes the response properties and spectral sensitivities of conesfrom the retina of the ground squirrel. It is shown that these cells have differentkinetics than those of primate cones. The spectral sensitivity curves are similar inshape to one another but have a somewhat different form than those of primatecones, thus not obeying the invariant spectral form described by Mansfield (1985).

METHODS

AnimalsGolden-mantled ground squirrels are native to the mountain ranges of the western United States

and southwestern Canada. Animals used in this study were captured during summer months in theSierra mountains of northern California. Squirrels were housed individually in cages containing anest box and cotton nesting material, and given ad libitum access to food and water.

PreparationAnimals were dark adapted for at least 1 h before being killed with an overdose of sodium

pentobarbitone, or by carbon dioxide asphyxiation. The eyes were removed and hemisected underdim red light, and all further manipulations were performed under infra-red light with the aid ofan infra-red-to-visible image converter (FJW Industries). The eyecup was divided into quartersand retinal pieces were isolated with a fine forceps into tissue culture medium (Liebowitz's L-15,Gibco) supplemented with vitamins and amino acids (BME mixture, Gibco). Pieces of isolatedretina were chopped and placed in the recording chamber, where they were continuously perfusedwith an oxygenated Ringer solution warmed to 37 'C. The superfusate contained the followingsalts (in mM): NaCl, 120; KCl, 3-6; NaHCO3, 20; MgCl2, 1-2; CaCl2, 2-4; buffered to pH 7-4 with 5%CO2 gas and 3 mM-HEPES buffer. The Ringer solution also contained vitamin and amino acidsupplements and glucose (10 mM). One rod spectral sensitivity curve was determined in anexperiment in which sulphate salts completely replaced chloride in the Ringer solution; sulphatedid not effect the rod spectrum.The outer-to-inner segment connection was quite fragile; often only a small number of cones

retained their outer segments through the isolation and chopping procedure. Almost all cells withouter segments responded to light; however, the largest photocurrents were routinely recordedfrom retinal pieces in which almost all neighbouring cones had long (6-8,tm), straight outersegments.

Electrical recording and light stimuliThe optical and recording apparatus have already been described (Baylor et al. 1979 a, 1984). The

addition of a pressure transducer (Omega Engineering Inc.) in parallel with the suction electrodeproved useful in stabilizing a cell once the outer segment had been drawn into the pipette. The cell'sphotocurrent and the stimulus were digitally recorded on video tape (Neuro Data Instruments)during the experiments and later analysed with a PDP 11/73 computer (Indec). The light monitoroutput and cell photocurrent were filtered identically to avoid introducing spurious responsedelays.

T. W. KRAFT200

RESPONSES IN SQUIRREL PHOTORECEPTORS

Wavelength was varied by interference filters with half-widths of about 10 nm (Ditric, 3 cavity).The light was attenuated by inconel neutral density filters (Bausch and Lomb). The power outputof the stimulating lamp through each of the interference filters was measured daily with aradiometer (United Detector Technologies). Details of the calibration of the interference andneutral density filters were as described by Baylor et al. (1984). Because of the relatively highsensitivity found at short wavelengths three of the interference filters (381, 402 and 421 nm) wererecalibrated at the end of the experiments, and found to be unchanged from previous calibrations.All stimuli consisted of unpolarized light.The shutter was driven by a powerful stepping motor (Tormax 20-010, IMC Magnetics Corp.)

controlled by an interval generator (WPI model 830). Photomultiplier measurements showed themaximum variation in the flash duration to be about 50 /us, or less than 05%.

Flash and step responsesA simple test of linearity is to observe the scaling of the cell's flash response amplitude with the

flash strength. A more severe test that reveals slowly developing non-linearities was described byBaylor & Hodgkin (1973); assuming linearity, the response to a step of light delivered at t = 0,R.(t), is predicted by the scaled integral of the response to a flash, R,(t)

RIM -= Rf(t) dt, (1)is ifAtj

where I. and If are the step and flash intensities respectively and At is the duration of the flash.

Collecting areaThe effective collecting area (Baylor & Hodgkin, 1973) of a cone can be calculated from the

measured dimensions of the outer segment, assuming values for the quantum efficiency ofisomerization, Qisom, and the specific pigment density, a. From Baylor, Lamb & Yau (1979b)

A = VoS Qisomf2-303a, (2)

where V.. is the volume of the outer segment, and f, is a factor that allows for the polarization ofthe stimulating light. The cone outer segment dimensions observed with the infra-red viewingsystem, 6-9 /tm length, 2 #um tip diameter and 3 4am base diameter, agreed with previous anatomicstudies (Jacobs et al. 1976). Using typical values for outer segment dimensions V,. is about 41 jum3.Qisom was assumed to have a value similar to rhodopsin, 0-67 (Dartnall, 1972). Micro-spectrophotometric measurements of pigment density in rods and cones reveal a to be near0016 /sm' (Rodieck, 1973; Harosi, 1975; Bowmaker, Dartnall & Mollon, 1980). Assuming adichroic ratio of 4, f is 0-63 for unpolarized light, and the resulting estimate for the collecting areais 0-64 fm2.

Spectral sensitivitySpectral sensitivity was measured relative to the sensitivity at 500 nm, using diffuse illumination

(300,um spot) incident transverse to the long axis of the outer segment. With a path length of2-3,um and assuming a specific pigment density of 0-016 gm-1 pigment self-screening can beignored and the measured sensitivity should be proportional to the probability of photonabsorption. Measurements at the standard wavelength were repeated frequently to avoid errorsassociated with changes in the physiological state of the cell. Spectra from individual cells werenormalized by shifting them on the logarithmic ordinate scale so that the mean log sensitivitieswere equated. Other details of the method have already been published (Baylor et al. 1984,1987).Polynomial expressions for pigment nomograms (Dawis, 1981) were used to estimate the

wavelength of maximum sensitivity (Amax) for the green-sensitive cones and the rods. The best fitwas found by minimizing the sum of the squared errors. The blue-sensitive cone data were fittedby a cubic spline routine (Mathsoft Inc.) and Amax estimated by interpolation.

201

T. W. KRAFT

RESULTS

Flash responsesFigure 1 shows a family of superimposed photocurrents evoked by flashes of

increasing strength in a green-sensitive cone; the inset is a schematic diagram of therecording procedure. The ordinate is the change of outer segment current from thelevel in darkness, and the abscissa is time from the centre of the stimuli. Brief flashesof dim light produced photocurrents that scaled linearly with flash strength (see

ios

60

pA

40

20

0 05 010 015 0-20s

Fig. 1. Family of superimposed responses from a green-sensitive cone to 11 ms flashes ofincreasing strength. Changes in dark current plotted as a function of time after the flash.Linear range responses rose to a peak in roughly 20 ms; saturating response was 67 pA.Each trace is the average of two to twenty-three sweeps. The upper trace is a single sweep.Flash intensities increased in nominal factors of two between 978 and 5 07 x 105photons 4um2. Flash monitor shown below current traces. Bandwidth 0-100 Hz,temperature 36 'C. i,., outer segment current.

below). Stronger flashes produced larger responses; the two brightest flashes shut offthe dark current entirely, producing a saturating photocurrent. Responses belowabout 15 pA reached a peak in about 20 ms in the records of Fig. 1, and the maximumresponse amplitude was 67 pA. The average saturating photocurrent was about 30pA (n = 10) in the better recordings.The photocurrent grew with flash strength according to the exponential saturation

function described by Lamb, McNaughton & Yau (1981) as shown in Fig. 2. Thenormalized peak photocurrent amplitude is plotted against the scaled flash strength.On the logarithmic abscissa the flash strengths have been multiplied by the constant

202

RESPONSES IN SQUIRREL PHOTORECEPTORS

k, the reciprocal of the flash strength that gave a response 0-63 of maximum. Thesmooth curve was drawn according to the expression

rirmax = 1-e-ki, (3)

where r is the peak response amplitude, rmax is the maximum response, k is theproportionality constant characteristic of the cell, and i is the stimulus strength. Theproportionality constant, k, is related to the half-saturating stimulus strength, i1, byk= (ln2)/ii.

2~~~~~~~~~~~~1~~~~~~~~~~~~~~~CA --A

r V- 0-5-rmax

0*01 0o1 1 10 100ki

Fig. 2. Intensity-response relations for seven cells. Peak response amplitude relative tomaximum value is plotted against the logarithm of the normalized photon density.Smooth curve drawn according to the exponential saturation function (eqn (3)).

The integration time of a cell is given by (Baylor & Hodgkin, 1973)00

ti = Rf(t)dt/Rpeak, (4)

where Rf(t) is the response to a brief flash of light at t = 0 and Rpeak is the peakamplitude of the flash response. The average integration time, (ti), for eleven coneswas 30 ms. The time to peak and integration time for a squirrel rod were 150 and209 ms respectively.The time to peak for blue-sensitive cone responses, 30 + 9 ms (mean + S.D., n = 7),

was similar to that measured for green-sensitive cones, 27+8 ms (n = 22). Thesepopulations were not significantly different based on a Wilcoxon rank-sum statistic(Brown & Hollander, 1977). For both cell types the fastest linear responses indarkness reached a peak in 20-22 ms (flash duration 11 ms), which is equivalent toan impulse response with a time to peak of 14 ms (see below).At the respective peak wavelengths (Amax), the flash sensitivity in darkness, SD,

and flash strength required for half saturating response, ii, were also indistinguishablefor the two types of cones. The greatest sensitivity to brief flashes in the dark was3-8 x 10-3 pA photon-',Um2, as measured in cells with the largest saturating

203

responses. The average sensitivity was 1P7 + 09 x 10-3 pA photon-' um2 (n = 16).The half-saturating intensity, 10900 + 2200 photons /tm-2 (n = 11), was less variablefrom cell to cell. This finding is consistent with the idea that in cells with a reduceddark current a portion of the outer segment was damaged or missing, the remainingouter segment being normal.

Impulse responseA simple quantitative description of the response to a brief flash is given by the

Baylor, Hodgkin & Lamb (1974) filter model. Their equation for the impulse responseto a series of n low pass filters with equal time constants written as in Baylor et al.(1984) is,

r*(t) = iSD{t* e(-t*)}nl1 (5)

where r*(t) is a linear range response, i is the flash strength, and t* is time after theflash normalized by the time to the peak of the response (t* = t/tp). Figure 3A showsthe averaged response to a dim flash of light (continuous line), and the impulseresponse predicted by eqn (5) with n 5, and tp = 22 ms (dotted curve).

Because the time to the peak of the response was short relative to the stimulusduration (22 vs. 11 ms), the stimulus was not truly impulse-like. To determine theimpulse response the stimulus and response were deconvoluted using fast Fourieranalysis. The cell response in Fig. 3A was modelled by eqn (5), and the stimulusrepresented by a square wave with duration 11-2 ms. The deconvolution is shown inFig. 3B (continuous line). The time to the peak of the predicted impulse response was16 ms. The shape of the curve was reasonably fitted by eqn (5) with n = 3 (Fig. 3B,dotted curve), except that the falling phase of the derived impulse response wasfaster than that predicted by eqn (5), this may indicate the existence of additionaldelays.

After correcting for the stimulus duration the average linear range cone responsesreached a peak in 21 ms; the fastest corrected time to peak was 14 ms.

Differentiation of the response to a step of light was also used to obtain an estimateof the impulse response of the cone (see Methods). This technique, applied toresponses from two additional cells also predicted impulse responses with a time topeak of about 16 ms.

LinearityFigure 4 demonstrates the linearity of flash responses that were below 20% of the

maximum response amplitude.The current traces show a cell's response to a brief flash (upper) and a step of light.

The dotted curve shows the step response predicted from the integral of the flashresponse (eqn (1), see Methods). These responses were from a relatively slow cellwhere the brief flash was a good approximation to an impulse. A slight undershootis evident in the flash response and a more dramatic dip is present in the response tothe 800 ms stimulus. Linearity was observed in similar tests of superposition on nineother cells.

204 T. W. KRAFT

RESPONSES IN SQUIRREL PHOTORECEPTORS

0.4 pA2

o 5l0 lOm0 50 100 ms

B~~~~~~~ a.

0 50 100 ms

Fig. 3. A, averaged responses to dim flashes of diffuse light (continuous curve, n = 185).Dotted curve drawn according to the five-stage filter model of eqn (5), T = 5-5 ms. Stimuliwere 11-2 ms flashes of 560 nm light, 1540 photons ,tm 2. Variance measurements from thesame series of flash responses are below the photocurrent tracing. Flash monitor shownbelow current traces. B, impulse response of the cell predicted from the deconvolution (seetext); the dotted curve shows the impulse response for a three-stage filter model of eqn(5). Bandwidth 0-150 Hz, temperature 35 'C.

Single photon responseAssuming a Poisson distribution of events consisting of single photon absorptions

giving responses of constant amplitude, the ensemble variance of the responseamplitudes (0-2) to a series of dim flashes should be related to the mean responseamplitude, ,t, by

O2 = aa, (6)

where a is the amplitude of the single photon response. The ratio of the variance tothe mean provides an upper limit for the amplitude of the single photon responsebecause several factors can increase the measured variance: (1) desensitizationduring recording, (2) changes in the seal resistance or length of outer segment in thesuction electrode, (3) drift in the dark current of amplifier and (4) real variation in

205

the amplitude of the single photon response. Many trials were required to obtainreliable estimates of the variance. The lower trace of Fig. 3A shows the currentvariance. The variance measured at the peak of the photocurrent was elevated by0-35 pA2, roughly a 40% increase. Thus for the data shown in Fig. 3A the singlephoton response amplitude was calculated to be 85 fA. Similar results from threeother cones gave values of about 80 fA.

2

pALo

5

pA rI l

0 0-5 1-0s

Fig. 4. Demonstration of linearity for small responses in a blue-sensitive cone. Theresponse to an 11 ms flash (upper trace) was used to predict the cell's response to an800 ms step (lower trace). The prediction is given by the dotted curve (see text forexplanation). Bandwidth 0-150 Hz, temperature 36 'C.

A second method of calculating the single photon response amplitude makes useof the relation,

a = ,zz/Ai, (7)

where ,t is the mean amplitude of the response to a brief flash, A is the collecting areaof the cell (see Methods) and i is the flash strength in photons jtm-2. Results fromthree cells where the specific dimensions of the outer segment were known gavevalues for a of about 7 fA.

Background lightThe effects of background light on flash responses are illustrated in Fig. 5. In Fig.

5 A flash responses recorded in the dark and with three background lights have beensuperimposed after scaling the amplitudes by the stimulus strengths. The uppermostcurve was obtained in the dark. Background light speeded the recovery after a flash,reduced the integration time, and made the response more symmetrical. Parametersof the three responses in background light shown in Fig. 5A are given by the firstthree filled circles of Fig. 5B and C. The time to the peak of the flash response (tp)

206 T. W. KRAFT

RESPONSES IN SQUIRREL PHOTORECEPTORS

1o

I

0 0-1 02s

log /b4

B 1-0

tp

tD

0-5

4_10

tj

F-5

5 6

* 0

0

0

O0

0

0 0

0

0

0 o 0 0 0

00

Fig. 5. Background light effects. A, superimposed responses to brief flashes in darkness(uppermost trace) and with background light present. Responses are scaled by theirstimulus strengths, 3400 and 6900 photons ,um-2 at 420 nm for the two upper and twolower traces, respectively. B and C, the time to the peak amplitude of the flash response

(t,) and integration time (t1) relative to their values in darkness are plotted for two cellsagainst the log of the background intensity (Ib). Bandwidth 0-150 Hz, temperature 36 'C.SF and SD are, respectively, the flash sensitivity in background light and in darkness.

and the cells' integration time (ti) were slightly reduced by background light. Timeto peak and integration times have been corrected for the duration of the flash. Theflash sensitivity was reduced by background light, but because there was incompleteand inconsistent recovery following background illumination no further conclusionswere drawn.

During recording, about one-third of the cells developed a small undershoot in therecovery phase of the response (e.g. Fig. 4). Background light did not produce or

enhance the undershoot.

Spectral sensitivities

Spectral sensitivity functions were measured in fourteen green-sensitive cones, sixblue-sensitive cones, and three rods from six animals. Action spectra were measuredby determining the flash sensitivity at the test wavelength relative to that at thereference wavelength (500 nm). Averaged data for the cones are plotted in Fig. 6.Nomogram fits to the spectral sensitivity functions (Dawis, 1981) were used to

A

SFSDF

207

rT

estimate the wavelength of maximum sensitivity for the green-sensitive cone androd. Data from green-sensitive cones were well fitted by a Dartnall nomogram withthe wavelength of maximum sensitivity (Amax) at 517 nm. The rod data were fittedby a Dartnall nomogram with Amax at 501 nm. The fit of the blue-sensitive conespectrum by the short-wavelength nomogram suggested by Liebman & Entine(1968) was quite poor, underestimating the cone sensitivity below 430 nm. A cubicspline fit to the data produced a curve quite like the one drawn by eye in Fig. 6;interpolation of that curve provided an estimate of the maximum sensitivity at

Wave number (m-')

0

-1

logs-2

-3 _-

-4

2*6 2*4 2*2 2*0 1*8 16 14I

400IlI I l

500 600 700

Wavelength (nm)

Fig. 6. Average normalized spectral sensitivity data for six blue-sensitive (@) andfourteen green-sensitive (0) cones from five golden-mantled ground squirrels. Continuouscurves drawn by eye. Dashed curves are corrected for strong short-wavelength absorptionby the lens (see text). S, relative sensitivity.

437 nm. The spectral bandwidth at half-height (WI) for the green-sensitive cone was

0-46 /tm-', and could not be measured for the blue-sensitive cone because of the highsensitivity (absorbance) at short wavelengths.The lenses ofmany ground squirrels contain a yellow pigment with extraordinarily

strong absorption at short wavelengths (Walls, 1942; Yolton, Yolton, Renz &Jacobs, 1974), the optical density being as high as 20 at 380 nm (Cooper & Robson,1969). The absorption of the lens should shift the effective sensitivity maxima of theblue- and green-sensitive cones in the intact eye to about 463 and 528 nmrespectively (dashed curves Fig. 6).

T

I

208 T. W. KRAFT

RESPONSES IN SQUIRREL PHOTORECEPTORS

Spectral formTrends in the shapes of spectral sensitivity curves from many species have led to

speculations on relations between the shape and peak wavelength (Greenberg, Honig& Ebrey, 1975; Ebrey & Honig, 1977; Bowmaker et al. 1980; Mansfield, 1985). Figure7 plots the normalized log sensitivity of the squirrel photoreceptors as a function oflog wave number; the symbols for the cones are the same as those in Fig. 6. A portionof the rod spectrum is shown by the open triangles. The spectra of the rod and blue-sensitive cone have been shifted along the log wave number axis to minimizedifferences with the green-sensitive cone spectrum. The continuous curve is the sixth-

0 -

0A~~~~~~~A

0

logS

-26

-4

l04 03 0-2

log wave number (pm-1)

Fig. 7. Spectral sensitivity curves plotted on a log wave number axis. Data for the blue-sensitive cones (-) and rods (A) have been shifted laterally for a best fit, by eye, to thecurve for the green-sensitive cones (0). Continuous curve is the sixth-order polynomialfrom Baylor et al. (1987); the maximum has been shifted to coincide with that of thesquirrel cones. The dashed curve is the same polynomial shifted to fit the descending limbof the spectral sensitivity curve. S, relative sensitivity.

order polynomial fitted by Baylor et al. (1987) to macaque cone spectra, shiftedlaterally to match the peak sensitivities for the squirrel cones. The shape of thesquirrel photoreceptor spectra are similar to one another but do not fit the generalshape of the monkey cone spectra. The dashed curve is the Baylor et al. polynomialshifted for a best fit by eye to the long-wavelength end of the squirrel spectra. Thedescending limb of both monkey and squirrel spectra coincide nicely at the long-wavelength end, but at the expense of a poor fit for the short-wavelength end of thespectra.

Spectral univarianceThe waveform of the response for all photoreceptors did not depend on the

wavelength of the stimulating light, as predicted by the univariance principle (Naka

209

& Rushton, 1966). As an additional test of univariance, wavelength sensitivity datafrom a green-sensitive cone were analysed at two times after the flash (25 and 130ms). The two spectral sensitivity curves derived were indistinguishable.

DISCUSSION

Photocurrent kinetics and sensitivityOne of the most striking characteristics of the squirrel cone responses was their

speed, most cells' responses had a corrected time to peak of about 20 ms. Salamanderand turtle cone photocurrents (Schnapf & MeBurney, 1980) reach peak in 100-200ms at room temperature. Assuming a Qlo of 2-7 (Baylor, Mathews & Yau, 1983), thesetimes would decrease to 23-46 ms on raising the temperature from 21 to 37 'C.Primate cone responses typically reach peak at 50-100 ms (Baylor et al. 1987;Schnapf et al. 1987).

Squirrel cone responses also lacked the undershoot characteristic of primate coneresponses. The ionic mechanisms responsible for the undershoot are unclear. Alteringthe internal calcium or calcium buffering capacity have produced undershoots inrods (Lamb, Matthews & Torre, 1985; Sather, Rispoli & Detwiler, 1988), presumablydue to calcium's ability to regulate guanylate cyclase activity (Pepe, Panfoli &Cugnoli, 1986; Koch & Stryer, 1988; Rispoli, Sather & Detwiler, 1988).The counterpart to the increased speed of the squirrel cone photocurrent is a low

sensitivity relative to other cones studied. The flash sensitivity in darkness in squirrelcones was five to ten times less than that of monkey and human cones (Baylor et al.1987; Schnapf et al. 1987). The flash strength required for half-maximal response(i!) in squirrel cones was correspondingly greater than that in primate cones.

Blue- and green-sensitive cones had indistinguishable kinetics and sensitivities.This finding is in agreement with similar data obtained from monkey cones (Schnapf,Nunn & Baylor, 1985), and supported by electroretinogram recordings from squirrelretinas which isolated the blue- and green-sensitive cone mechanisms (Crognale &Jacobs, 1988).

Single photon responseThe portion of channels closed by a single photoisomerization in turtle cones has

been estimated as 0-16% (Baylor et al. 1974) and 007% (Schnapf & McBurney,1980). For a conservative estimate in the squirrel cone consider a 50 pA dark currentand a single photon response of 10 fA. This figure represents a reduction of only0-02% of the dark current. Assuming a uniform current density through an outersegment with about 200 discs (Long & Fisher, 1983), this current reductionrepresents the closure of channels over only about ' of a disc. Based on thesegeometrical considerations the spread of excitation for a single activated photo-pigment would cover only a 055tm diameter circular area on one side of a disc.

Colour vision in squirrelsBehavioural observations on the golden-mantled ground squirrel indicated cone

sensitivity peaks of 440 and 525 nm (Jacobs, 1978). Ganglion cell recordings fromintact eyes of Mexican and thirteen-lined ground squirrels (Michael, 1968; Gur &

210 T. W. KRAFT

RESPONSES IN SQUIRREL PHOTORECEPTORS

Purple, 1978) gave estimated spectral peaks at 460 and 525 nm, very near thoseobtained here after correction for lens absorption (463 and 528 nm). Raisanen &Dawis (1983) reported a peak of 516 nm for the green-sensitive cone frommeasurements of the mass receptor potential (PIII). Jacobs et al. (1985) confirmedthat peak, and estimated the blue-sensitive cone peak at 437 nm. These recentstudies of cone spectral sensitivity correspond well with the action spectra reportedhere.The proportion of cones identified as blue sensitive in this study, seven of thirty-

eight (18%), was somewhat higher than previous estimates of 7-10% in groundsquirrels (Long & Fisher, 1983; Ahnelt, 1985). The sample size is relatively small, butto date thirteen blue-sensitive cones have been identified in a population of ninety-four cells from thirteen animals.

Spectral formThe shapes of the two squirrel cone spectra are similar to one another when plotted

on a log wave number axis, in agreement with the finding that pigment bandwidthsare inversely related to their peak wavelength (Ebrey & Honig, 1977; Mansfield,1985). The shapes of the squirrel spectra do not, however, conform to the idea of aninvariant spectral form for all vertebrate pigments (Mansfield, 1985; MacNichol,1986). The Baylor et al. (1987) template provides a very good fit to squirrel spectraat long wavelengths (dashed curve in Fig. 7) but falls below the measured sensitivitiesat wavelengths shorter than the peak. A possible explanation for the discrepancy atshort wavelengths could be an unusually prominent cis-band in squirrel pigments. Inany case these results suggest that the form of the spectral sensitivity curve is notgeneral across all mammalian species.

I thank Dr Denis Baylor for advice, Drs Clint Makino and Jeff Karpen for reviewing themanuscript, Carolyn Radeke and Dr Craig Heller for help in obtaining squirrel tissue, and MrR. Schneeveis and Mr W. Morris for technical support. This work was supported by grants EY05750 and EY 05956 from the National Institutes of Health, USPHS.

REFERENCES

AHNELT, P. K. (1985). Characterization of the color related receptor mosaic in the ground squirrelretina. Vision Research 25, 1557-1567.

BAYLOR, D. A. & HODGKIN, A. L. (1973). The electrical response of turtle cones to flashes and stepsof light. Journal of Physiology 242, 685-727.

BAYLOR, D. A., HODGKIN, A. L. & LAMB, T. D. (1974). Detection and resolution of visual stimuliby turtle photoreceptors. Journal of Physiology 234, 163-198.

BAYLOR, D. A., LAMB, T. D. & YAU, K.-W. (1979a). The membrane current of single rod outersegments. Journal of Physiology 288, 589-611.

BAYLOR, D. A., LAMB, T. D. & YAU, K.-W. (1979b). Responses of retinal rods to single photons.Journal of Physiology 288, 612-634.

BAYLOR, D. A., MATHEWS, G. & YAU, K.-W. (1983). Temperature effects on the membrane currentof retinal rods of the toad. Journal of Physiology 337, 723-734.

BAYLOR, D. A., NUNN, B. J. & SCHNAPF, J. L. (1984). The photocurrent, noise and spectralsensitivity of rods of the monkey Macaca fascicularis. Journal of Physiology 357, 575-607.

BAYLOR, D. A., NUNN, B. J. & SCHNAPF, J. L. (1987). Spectral sensitivity of cones of the monkey.Macaca fascicularis. Journal of Physiology 390, 145-160.

211

BOWMAKER, J. K., DARTNALL, H. J. A. & MOLLON, ,J. D. (1980). Microspectrophotometricdemonstration of four classes of photoreceptor in an old world primate, Macaca fa.scicularis.Journal of Physiology 298, 131-143.

BROWN JR, B. W. & HOLLANDER, M. (1977). Statistics: A Biom-tedical Introduction. New York: JohnWiley & Sons.

COOPER, G. F. & ROBSON, J. G. (1969). The yellow colour of the lens of the grey squirrel (Sciuruscarolinensis leucotis). Journal of Physiology 203, 403-410.

CRESCITELLI, F. & POLLACK, J. D.. (1972). Dichromacv in the antelope ground squirrel. VisionResearch 12, 1553-1586.

CROGNALE, M. & JACOBS, G. H. (1988). Temporal properties of the short-wavelength conemechanism: Comparison of receptor and postreceptor signals in the ground squirrel. V'isionResearch (in the Press).

DARTNALL, H. J. A. (1972). Handbook of Sensory Physiology VTT/I: Photochemistry of Vision. NewYork: Springer-Verlag.

DAWIS, S. M. (1981). Polynomial expressions of pigment nomograms. Vision Research 21, 1427-1430.

EBREY, T. G. & HONIG, B. (1977). New wavelength dependent visual pigment nomograms. VisionResearch 17, 147-151.

GREENBERG, A. D., HONIG, B. & EBREY, T. G. (1975). Wavelength depelndence of the bandwidthsof visual pigment spectra. Nature 257, 823-824.

GUTR, M. & PURPLE, R. L. (1978). Retinal ganglion cell activity in the ground squirrel underhalothane anesthesia. Vision Research 18, 1-14.

HAROSI, F. I. (1975). Absorption spectra and linear dichroism of some amphibian photoreceptors.Journal of General Physiology 66, 357-382.

JACOBS, G. H. (1978). Spectral sensitivity and colour vision in the ground-dwelling sciurids:Results form golden-mantled ground squirrels and comparisons for five species. Animal Behavior26, 409-421.

JACOBS, G. H., FISHER, S. K., ANDERSON, D. H. & SILVERMAN, M. S. (1976). Scotopic and photopicvision in the California ground squirrel: Physiological and anatomical evidence. Journal oJComparative Neurology 165, 209-228.

JACOBS, G. H., NEITZ, J. & CROGNALE, M. (1985). Spectral sensitivity of ground squirrel cones

measured with ERG flicker photometry. Journal of Comparative Physiology A 156, 503-509.KOCH, K.-W. & STRYER, L. (1988). Regulation of bovine rod outer segment guanylate cyclase by

calcium. Biophysical Journal 53, 388a (abstract).LAMB, T. D., MCNAUGHTON. P. A. & YAU, K.-W. (1981). Spatial spread of activation and

desensitization in toad rod outer segments. Joaurnal of Physiology 319. 463-496.LAMB, T. D., MATTHEWS, H. R. & TORRE. V. (1985). Incorporation of calcium buffers into

salamander retinal rods: A rejection of the calcium hypothesis of phototransduction. Journal ofPhysiology 372, 315-349.

LIEBMAN, B. & ENTINE, G. (1968). Visual pigments of frog and tadpole. lVision Research 8, 761-776.

LONG, K. 0. & FISHER, S. K. (1983). The distribution of photoreceptors and ganglion cells in theCalifornia ground squirrel, Spermophilus beecheyi. Journal of Comparative Neurology 221, 329-340.

MACNICHOL JR, E. F. (1986). A unifying presentation of photopigment spectra. Vision Research26, 1543-1556.

MANSFIELD, R. J. W. (1985). Primate photopigments and cone mechanisms. In The Visual Systenm.ed. FEIN, A. & LEVINE, J. S., pp. 89-106. New York: A. R. Liss.

MICHAEL, C. R. (1968). Receptive fields of single optic nerve fibers in a mammal with an all-coneretina. III: Opponent color units. Journal of Neurophysiology 31, 268-282.

NAKA, K.-I. & RUSHTON, W. A. H. (1966). S-potentials from colour units in the retina of fish(Cyprinidae). Journal of Physiology 185, 536-555.

PEPE, I. M., PANFOLI, I. & CUGNOLI, C. (1986). Gxuanylate cyclase in rod outer segments of the toadretina. Federation of European Biochemical Society 203, 73-76.

RAISANEN, J. & DAWIS, S. M. (1983). A reweighting of receptor mechanisms in the ground squirrelretina: Pll and b-wave spectral sensitivity functions. Brain Research 270, 311-318.

RISPOLI, G., SATHER, W. A. & DETWILER, P. B. (1988). Effect of triphosphate nucleotides on the

T. W. KRAFT212

RESPONSES IN SQUIRREL PHOTORECEPTORS 213

response of detached rod outer segments to low external calcium. Biophysical Journal 53, 388a(abstract).

RODIECK, R. W. (1973). The Vertebrate Retina. San Francisco: W. H. Freeman.SATHER, W. A., RISPOLI, G. & DETWILER, P. B. (1988). Effect of calcium on light adaptation in

detached gecko rod outer segments. Biophysical Journal 53, 390a (abstract).SCHNAPF, J. L.. KRAFT, T. WV. & BAYLOR. D. A. (1987). Spectral sensitivity of human cone

photoreceptors. Nature 325, 439-441.SCHNAPF, J. L. & McBURNEY. R. N. (1980). Light-induced changes in membrane current in cone

outer segments of tiger salamander and turtle. Nature 287, 239-241.SCHNAPF, J. L., NUNN, B. J. & BAYLOR, D. A. (1985). Visual transduction in monkev cones.

Investigative Ophthalmology 26, suppl., 194.TANSLEY, K., COPENHAVER, R. M. & GUNKEL, R. D. (1961). Spectral sensitivity curves of diurnal

squirrels. Vision Research 1, 154-165.WALLS, G. L. (1942). The Vertebrate Eye and Its Adaptive Radiation. Bloomfield Hills, MI, U.S.A.:

Cranbrook Institute of Science.WVEST, R. W. & DOWLING, J. E. (1975). Anatomical evidence for cone and rod-like receptors in thegray squirrel. ground squirrel and prairie dog retinas. Journal of Comparative Neurology 159,439-460.

YOLTON, R. L., YOLTON, D. P., RENZ, J. & JACOBS, G. H. (1974). Preretinal absorbance in sciurideyes. Journal of Mammalogy 55, 14-20.